What Is Template Switching and How Does It Work?

Template switching is what happens when a polymerase, the molecular machine copying nucleic acids, jumps from one template strand to another mid-synthesis and keeps going on the new template as if nothing happened. The result is a hybrid product stitched together from pieces of two different molecules. This seemingly simple event turns out to be everywhere in biology: retroviruses depend on it to replicate, coronaviruses use it to make their messenger RNAs, your own cells use it to rescue stalled DNA replication, and biotechnologists have harnessed it to build single-cell sequencing libraries. The mechanism and consequences vary depending on context, but the core idea is always the same: the polymerase lets go of one template and grabs another.

The Basic Mechanism in Retroviral Replication

The best-studied version of template switching happens during reverse transcription, the process retroviruses like HIV use to convert their RNA genome into DNA. A retrovirus carries two copies of its RNA genome inside each viral particle, and the enzyme reverse transcriptase copies them into DNA. But reverse transcriptase cannot simply read one RNA molecule end to end and produce a complete DNA copy. Instead, it must perform at least two obligatory “strand transfers,” each of which is a template switch, just to finish a single round of DNA synthesis.

During the first of these, reverse transcriptase starts copying near the 5′ end of one RNA molecule and synthesizes a short stretch of DNA. Then it jumps to the 3′ end of the same (or the other) RNA template, guided by a short repeated sequence called R that appears at both ends of the viral RNA. The newly made DNA pairs with this matching R sequence at the new location, and synthesis continues. A second obligatory strand transfer occurs later during plus-strand DNA synthesis, using another region of sequence complementarity. Without these jumps, the virus cannot produce a complete double-stranded DNA copy of its genome.

In the lab, the efficiency of this switching can be remarkably high. Early experiments with RNA templates and purified reverse transcriptase showed that up to about 83% of synthesis reactions underwent template switching. The enzyme’s built-in RNase H activity, which degrades RNA in RNA-DNA hybrids, helps by chewing up the old template as the DNA is being made, freeing the new DNA strand to find the matching sequence on the other template. When RNase H activity was removed, switching still happened but at a lower rate. And when the template was switched from RNA to DNA, template switching persisted, largely independent of RNase H, because there was no RNA left to degrade.¹

How Template Switching Drives Viral Recombination

The obligatory strand transfers described above are essential housekeeping for the virus. But reverse transcriptase also switches templates at internal sites along the genome, and this is where things get interesting for viral evolution. Because each retroviral particle packages two RNA molecules, reverse transcriptase can jump between them during copying. When it does this at internal positions, the resulting DNA is a mosaic of sequences from both parental RNAs. This is copy-choice recombination: the enzyme “chooses” a new template copy partway through synthesis, producing recombinant offspring in a single replication cycle.²

The consequences ripple outward. Intermolecular switching between the two copackaged genomes generates recombinant viruses with reshuffled genetic material, accelerating the evolution of drug resistance or immune evasion. Intramolecular switching, where the enzyme jumps within the same RNA molecule, can produce defective genomes with internal deletions.³ HIV is especially prone to recombination; the many circulating recombinant forms of HIV-1 found worldwide are a direct consequence of template switching during co-infection with two different viral strains.

Several factors influence where these internal switches happen. Regions of high sequence similarity between the two RNA copies make it easier for the new DNA strand to find a match and anneal. Viral proteins also play a role: the nucleocapsid protein, for example, helps by destabilizing RNA secondary structures that would otherwise block the new DNA from pairing with the acceptor template.⁴ Even relatively short stretches of repeated sequence can serve as landing pads; experiments with mutant HIV showed that repeats much shorter than the wild-type 97-nucleotide R region still supported efficient strand transfer.⁵

Coronavirus Transcription and Discontinuous Synthesis

Coronaviruses, including SARS-CoV-2, use a version of template switching that serves a different purpose from retroviral recombination. These viruses carry a single large RNA genome but need to produce a set of smaller messenger RNAs (called subgenomic RNAs) to express their structural and accessory proteins. To make these smaller RNAs, the viral replication complex performs a discontinuous synthesis step that is, at its core, a template switch.

During negative-strand RNA synthesis, the viral polymerase is copying the positive-sense genome when it encounters specific regulatory sequences called transcription-regulating sequences (TRS). At these points, the polymerase can jump from the body TRS to a matching leader TRS near the 5′ end of the genome. The result is a negative-strand RNA that fuses the leader sequence to the body of the gene, and when this gets copied back into positive-sense RNA, the cell gets a subgenomic messenger RNA with the leader attached.⁶ The extent of base-pairing between the body and leader TRS sequences regulates how efficiently the polymerase jumps, and multiple viral and host factors fine-tune the process.

Recent experimental work on both SARS-CoV-2 and another human coronavirus (OC43) has provided evidence that the viral replication complex synthesizes these subgenomic RNAs from the 3′ end of the genome moving in a 5′ to 3′ direction, encountering TRS sites in order. The subgenomic RNA encoding the nucleocapsid (N) protein, whose TRS is closest to the 3′ end, is produced most abundantly, while the spike (S) protein RNA, whose TRS is farther away, is produced in lesser amounts. This supports a “first-come, first-served” model for template switching during coronavirus transcription.⁷

Template Switching Beyond Viruses: Rescuing Stalled Replication Forks

Template switching is not limited to viruses. Your own cells use it as part of a DNA damage tolerance strategy during normal DNA replication. When a replication fork runs into a damaged base, a tightly bound protein, or an unusual DNA structure, it can stall. If the cell cannot immediately repair the damage, it faces a problem: the copying machinery is stuck, and leaving it stuck risks chromosome breakage. Template switching offers one way out.

In this scenario, the newly synthesized strand at the stalled fork peels back and pairs with the undamaged, newly synthesized sister strand behind the fork. By using the sister strand as a temporary template, the replication machinery can extend past the position of the lesion without ever actually reading the damaged base. Once it has enough sequence copied from the sister strand, the new DNA can be re-annealed to the original template and synthesis continues. The damaged base remains unrepaired but is effectively bypassed, and because the sister strand was used as a template, no mutations are introduced. This is why it is often described as error-free damage tolerance, in contrast to translesion synthesis, where specialized but sloppy polymerases read through damage at the cost of frequent mistakes.⁸

Cells regulate the choice between translesion synthesis and template switching through modifications to a protein called PCNA, which acts as a sliding clamp for the replication machinery. Different types of ubiquitin tags on PCNA steer the fork toward one pathway or the other.⁹ Replication forks stall frequently in every dividing cell, thanks to encounters with all sorts of obstacles, and the ability to switch templates and then switch back is one of several strategies the cell uses to keep genomic information intact.¹⁰

When Template Switching Goes Wrong in Our Own Genomes

If error-free template switching is the cell’s careful workaround, things can get messy when the process goes off-script. Research in bacteria has shown that template switching during replication fork repair can contribute to mutational hotspots and genetic rearrangements, particularly when the fork encounters repeated sequences or secondary structures in the DNA.¹¹

In eukaryotes, when a replication fork collapses entirely and break-induced replication kicks in to restart it, the newly synthesized strand can switch between templates at positions where only tiny stretches of shared sequence exist. This process, called microhomology-mediated break-induced replication (MMBIR), is thought to be a major source of complex genomic rearrangements, including duplications, deletions, and translocations that underlie some human genetic diseases. A model proposes that under stress, when normal homologous recombination is unavailable, broken forks anneal at microhomologies with whatever single-stranded DNA is nearby, priming low-fidelity synthesis with multiple template switches and generating convoluted structural changes.¹²

Yeast experiments have demonstrated this transition directly. When normal break-induced replication is disrupted, cells fall back on microhomology-mediated repair that involves specialized polymerases capable of initiating DNA synthesis at sites with as few as zero to six matching nucleotides.¹³ The result can be genome rearrangements that look nothing like clean recombination events. Studies in fission yeast have further shown that when a stalled replication fork restarts through homologous recombination, the nascent strands can exchange templates between repeated sequences, producing inversions and other structural changes.¹⁴

Template Switching in RNA Virus Recombination More Broadly

Beyond retroviruses and coronaviruses, copy-choice recombination via template switching is widespread among RNA viruses. RNA-dependent RNA polymerases (RdRps), which copy RNA genomes in viruses that never go through a DNA stage, can also switch templates during elongation. In vitro experiments with RdRps from bovine viral diarrhea virus and several plant viruses showed that these enzymes produce detectable recombination products by jumping templates during active RNA synthesis. The frequency of switching was influenced by template sequence, RNA structure, and the availability of nucleotide building blocks.¹⁵

Enteroviruses, like poliovirus, use what is sometimes called forced-copy-choice recombination. If the polymerase encounters a break or damage in the RNA template, it falls off and then re-engages a different RNA molecule, stitching fragments together in the process. Recent structural and biochemical work suggests this mechanism may function as a last-resort strategy to rebuild a full-length genome when the original has been damaged by nucleases or chemical attack.¹⁶ The practical upshot is that RNA viruses generally are prolific recombiners, and template switching is a big part of why their genomes are so fluid.

How Biotechnology Exploits Template Switching

Scientists have turned template switching from a biological curiosity into a laboratory tool. The most prominent application is in single-cell RNA sequencing (RNA-seq) library preparation, where the natural behavior of certain reverse transcriptases is exploited to add adapter sequences to the ends of newly made cDNA.

Here is how it works: reverse transcriptases derived from Moloney Murine Leukemia Virus (MMLV) have an intrinsic tendency to add a few extra nucleotides, especially cytosines, at the 3′ end of cDNA when they reach the 5′ end of the RNA template. If an oligonucleotide with matching guanines on its 3′ end (called a template-switching oligonucleotide, or TSO) is floating in the reaction, the reverse transcriptase base-pairs the C overhang with the G’s on the TSO and continues synthesizing DNA using the TSO as its new template.¹⁷ The result is a cDNA molecule with a known adapter sequence seamlessly attached to its 5′ end, ready for amplification and sequencing. This approach underpins popular protocols like SMART-seq and its derivatives.

Biochemical studies have shown that the non-templated nucleotide addition and the template-switching step happen concurrently rather than sequentially; the two processes compete with each other in real time, and the timing of TSO addition affects the efficiency of the switch.¹⁸ Practical optimizations, such as choosing the right enzyme variant and reducing primer concentrations, have been used to make these protocols more sensitive and cheaper. One optimized single-cell RNA-seq method showed that reducing the template-switching oligo cost and enzyme amount had no measurable effect on cDNA yield or quality.¹⁹

A related development involves group II intron-encoded reverse transcriptases, which differ from retroviral enzymes in being able to switch templates efficiently even when there is little or no sequence similarity between donor and acceptor. This property has been adapted for both RNA-seq and DNA-seq adapter-addition protocols, offering an alternative to MMLV-based methods.²⁰

The Artifact Problem

The same template-switching behavior that makes reverse transcriptases useful in the lab also creates headaches for researchers who are not expecting it. When reverse transcriptase switches templates during routine cDNA synthesis from a complex mixture of messenger RNAs, it can produce chimeric cDNA molecules that look like real alternative transcripts but are pure artifact. A cDNA with an internal deletion might be mistaken for a genuinely spliced mRNA variant, leading to false conclusions about gene expression.²¹

A similar issue arises when working with long-read sequencing technologies that rely on reverse transcription. Template-switching artifacts can mimic alternative polyadenylation events, generating reads that appear to represent genuine sites where a transcript’s poly(A) tail was added, but are actually the result of the enzyme jumping to A-rich stretches elsewhere. Researchers have developed computational algorithms to distinguish these artifacts from real transcription end sites, and the results beat conventional filtering methods. In one comparison, the artifact-aware algorithm achieved a positive predictive value of about 75% in human samples compared with roughly 72% for the standard approach, and had a substantially higher negative predictive value.²²

PCR amplification of mixed templates introduces a related but distinct artifact. When PCR is run on a mixture of similar sequences, incomplete extension products from one template can anneal to a different template in subsequent cycles, creating chimeric amplicons. This kind of template switching during PCR tends to worsen in the later cycles of the reaction, when templates become abundant and primers become limiting.²¹ Awareness of these artifacts is critical in any experiment involving mixed or highly similar templates, from environmental microbiome surveys to cancer genomics.

Template Switching in Genome Editing

The concept has even been coopted for genome engineering. Prime editing, a relatively recent CRISPR-based technology, uses a reverse transcriptase fused to a Cas9 nickase to write new sequences directly into the genome. A newer extension of this approach, called template-jumping prime editing (TJ-PE), exploits template switching to insert large DNA sequences. In this system, the reverse transcriptase first synthesizes a stretch of DNA using the prime editing guide RNA as a template, then the newly made DNA strand switches to a second nicked site elsewhere on the genome, initiating a second round of synthesis. This two-step template jump allows insertions that would be too large for standard prime editing.²³ The approach has been used to insert sequences of hundreds of base pairs and even to rewrite entire exons in living cells, demonstrating that template switching can be deliberately engineered for therapeutic purposes.

Mobile Genetic Elements and the Evolutionary Angle

Template switching likely predates viruses and genome editing tools alike. Mobile genetic elements called group II introns, which are ancient self-splicing segments of RNA found in bacteria and organelles, encode their own reverse transcriptases. These enzymes are thought to be evolutionary ancestors of retroviral reverse transcriptases, and they template-switch with high efficiency, even between molecules with very little sequence similarity.²⁰

Structural studies have begun to reveal how these enzymes accomplish this feat. An X-ray crystal structure of a group II intron reverse transcriptase caught in the act of template switching showed the enzyme simultaneously bound to both an acceptor RNA and a donor RNA-DNA hybrid, with just a single-nucleotide 3′ DNA overhang bridging the two. This snapshot suggests the enzyme has an active site geometry that accommodates both the old and new templates at once, providing a structural explanation for how the switch happens so smoothly.²⁴

One hypothesis is that this template-switching ability helps mobile elements maintain their integrity over evolutionary time. If a host cell’s nucleases partially degrade a retroelement’s RNA before it can be reverse-transcribed, the reverse transcriptase could jump from the damaged end of one template to an intact copy and keep going, rescuing an otherwise doomed insertion event. The same logic applies to forced-copy-choice recombination in enteroviruses: the ability to stitch together RNA fragments may be less about generating diversity and more about surviving in hostile cellular environments where RNA is constantly under attack.